Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: The CADDementia challenge

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimer's disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n = 30) and optionally on data from other sources (e.g., the Alzheimer's Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org.

Gaussian processes autoencoder for dimensionality reduction

Learning low dimensional manifold from highly nonlinear data of high dimensionality has become increasingly important for discovering intrinsic representation that can be utilized for data visualization and preprocessing. The autoencoder is a powerful dimensionality reduction technique based on minimizing reconstruction error, and it has regained popularity because it has been efficiently used for greedy pretraining of deep neural networks. Compared to Neural Network (NN), the superiority of Gaussian Process (GP) has been shown in model inference, optimization and performance. GP has been successfully applied in nonlinear Dimensionality Reduction (DR) algorithms, such as Gaussian Process Latent Variable Model (GPLVM). In this paper we propose the Gaussian Processes Autoencoder Model (GPAM) for dimensionality reduction by extending the classic NN based autoencoder to GP based autoencoder. More interestingly, the novel model can also be viewed as back constrained GPLVM (BC-GPLVM) where the back constraint smooth function is represented by a GP. Experiments verify the performance of the newly proposed model.

Rapid preconditioning of data for accelerating convex hull algorithms

Given a dataset of two-dimensional points in the plane with integer coordinates, the method proposed reduces a set of n points down to a set of s points s ≤ n, such that the convex hull on the set of s points is the same as the convex hull of the original set of n points. The method is O(n). It helps any convex hull algorithm run faster. The empirical analysis of a practical case shows a percentage reduction in points of over 98%, that is reflected as a faster computation with a speedup factor of at least 4.

Asynchronous distributed parallelization of mobile network optimization algorithms

It has been years since the introduction of the Dynamic Network Optimization (DNO) concept, yet the DNO development is still at its infant stage, largely due to a lack of breakthrough in minimizing the lengthy optimization runtime. Our previous work, a distributed parallel solution, has achieved a significant speed gain. To cater for the increased optimization complexity pressed by the uptake of smartphones and tablets, however, this paper examines the potential areas for further improvement and presents a novel asynchronous distributed parallel design that minimizes the inter-process communications. The new approach is implemented and applied to real-life projects whose results demonstrate an augmented acceleration of 7.5 times on a 16-core distributed system compared to 6.1 of our previous solution. Moreover, there is no degradation in the optimization outcome. This is a solid sprint towards the realization of DNO.

Elephants in pyjamas: testing the weak central coherence account of autism spectrum disorders using a syntactic disambiguation task

According to the weak central coherence (CC) account individuals with autism spectrum disorders (ASD) exhibit enhanced local processing and weak part-whole integration. CC was investigated in the verbal domain. Adolescents, recruited using a 2 (ASD status) by 2 (language impairment status) design, completed an aural forced choice comprehension task involving syntactically ambiguous sentences. Half the picture targets depicted the least plausible interpretation, resulting in longer RTs across groups. These were assumed to reflect local processing. There was no ASD by plausibility interaction and consequently little evidence for weak CC in the verbal domain when conceptualised as enhanced local processing. Furthermore, there was little evidence that the processing of syntactically ambiguous sentences differed as a function of ASD or language-impairment status.

Conditioning of the weak-constraint variational data assimilation problem for numerical weather prediction

4-Dimensional Variational Data Assimilation (4DVAR) assimilates observations through the minimisation of a least-squares objective function, which is constrained by the model flow. We refer to 4DVAR as strong-constraint 4DVAR (sc4DVAR) in this thesis as it assumes the model is perfect. Relaxing this assumption gives rise to weak-constraint 4DVAR (wc4DVAR), leading to a different minimisation problem with more degrees of freedom. We consider two wc4DVAR formulations in this thesis, the model error formulation and state estimation formulation. The 4DVAR objective function is traditionally solved using gradient-based iterative methods. The principle method used in Numerical Weather Prediction today is the Gauss-Newton approach. This method introduces a linearised `inner-loop' objective function, which upon convergence, updates the solution of the non-linear `outer-loop' objective function. This requires many evaluations of the objective function and its gradient, which emphasises the importance of the Hessian. The eigenvalues and eigenvectors of the Hessian provide insight into the degree of convexity of the objective function, while also indicating the difficulty one may encounter while iterative solving 4DVAR. The condition number of the Hessian is an appropriate measure for the sensitivity of the problem to input data. The condition number can also indicate the rate of convergence and solution accuracy of the minimisation algorithm. This thesis investigates the sensitivity of the solution process minimising both wc4DVAR objective functions to the internal assimilation parameters composing the problem. We gain insight into these sensitivities by bounding the condition number of the Hessians of both objective functions. We also precondition the model error objective function and show improved convergence. We show that both formulations' sensitivities are related to error variance balance, assimilation window length and correlation length-scales using the bounds. We further demonstrate this through numerical experiments on the condition number and data assimilation experiments using linear and non-linear chaotic toy models.

Efficiency of evolutionary algorithms in water network pipe sizing

The pipe sizing of water networks via evolutionary algorithms is of great interest because it allows the selection of alternative economical solutions that meet a set of design requirements. However, available evolutionary methods are numerous, and methodologies to compare the performance of these methods beyond obtaining a minimal solution for a given problem are currently lacking. A methodology to compare algorithms based on an efficiency rate (E) is presented here and applied to the pipe-sizing problem of four medium-sized benchmark networks (Hanoi, New York Tunnel, GoYang and R-9 Joao Pessoa). E numerically determines the performance of a given algorithm while also considering the quality of the obtained solution and the required computational effort. From the wide range of available evolutionary algorithms, four algorithms were selected to implement the methodology: a PseudoGenetic Algorithm (PGA), Particle Swarm Optimization (PSO), a Harmony Search and a modified Shuffled Frog Leaping Algorithm (SFLA). After more than 500,000 simulations, a statistical analysis was performed based on the specific parameters each algorithm requires to operate, and finally, E was analyzed for each network and algorithm. The efficiency measure indicated that PGA is the most efficient algorithm for problems of greater complexity and that HS is the most efficient algorithm for less complex problems. However, the main contribution of this work is that the proposed efficiency ratio provides a neutral strategy to compare optimization algorithms and may be useful in the future to select the most appropriate algorithm for different types of optimization problems.

Weak polygyny in California sea lions and the potential for alternative mating tactics

Female aggregation and male territoriality are considered to be hallmarks of polygynous mating systems. The development of genetic parentage assignment has called into question the accuracy of behavioral traits in predicting true mating systems. In this study we use 14 microsatellite markers to explore the mating system of one of the most behaviorally polygynous species, the California sea lion (Zalophus californianus). We sampled a total of 158 female-pup pairs and 99 territorial males across two breeding rookeries (San Jorge and Los Islotes) in the Gulf of California, Mexico. Fathers could be identified for 30% of pups sampled at San Jorge across three breeding seasons and 15% of sampled pups at Los Islotes across two breeding seasons. Analysis of paternal relatedness between the pups for which no fathers were identified (sampled over four breeding seasons at San Jorge and two at Los Islotes) revealed that few pups were likely to share a father. Thirty-one percent of the sampled males on San Jorge and 15% of the sampled males on Los Islotes were assigned at least one paternity. With one exception, no male was identified as the father of more than two pups. Furthermore, at Los Islotes rookery there were significantly fewer pups assigned paternity than expected given the pool of sampled males (p<0.0001). Overall, we found considerably lower variation in male reproductive success than expected in a species that exhibits behavior associated with strongly polygynous mating. Low variation in male reproductive success may result from heightened mobility among receptive females in the Gulf of California, which reduces the ability of males to monopolize groups of females. Our results raise important questions regarding the adaptive role of territoriality and the potential for alternative mating tactics in this species.

Selecting algorithms for Earth observation of climate within the European Space Agency Climate Change initiative: introduction to a special issue

This special issue is focused on the assessment of algorithms for the observation of Earth’s climate from environ- mental satellites. Climate data records derived by remote sensing are increasingly a key source of insight into the workings of and changes in Earth’s climate system. Producers of data sets must devote considerable effort and expertise to maximise the true climate signals in their products and minimise effects of data processing choices and changing sensors. A key choice is the selection of algorithm(s) for classification and/or retrieval of the climate variable. Within the European Space Agency Climate Change Initiative, science teams undertook systematic assessment of algorithms for a range of essential climate variables. The papers in the special issue report some of these exercises (for ocean colour, aerosol, ozone, greenhouse gases, clouds, soil moisture, sea surface temper- ature and glaciers). The contributions show that assessment exercises must be designed with care, considering issues such as the relative importance of different aspects of data quality (accuracy, precision, stability, sensitivity, coverage, etc.), the availability and degree of independence of validation data and the limitations of validation in characterising some important aspects of data (such as long-term stability or spatial coherence). As well as re- quiring a significant investment of expertise and effort, systematic comparisons are found to be highly valuable. They reveal the relative strengths and weaknesses of different algorithmic approaches under different observa- tional contexts, and help ensure that scientific conclusions drawn from climate data records are not influenced by observational artifacts, but are robust.

Zero attracting recursive least squares algorithms

The l1-norm sparsity constraint is a widely used technique for constructing sparse models. In this contribution, two zero-attracting recursive least squares algorithms, referred to as ZA-RLS-I and ZA-RLS-II, are derived by employing the l1-norm of parameter vector constraint to facilitate the model sparsity. In order to achieve a closed-form solution, the l1-norm of the parameter vector is approximated by an adaptively weighted l2-norm, in which the weighting factors are set as the inversion of the associated l1-norm of parameter estimates that are readily available in the adaptive learning environment. ZA-RLS-II is computationally more efficient than ZA-RLS-I by exploiting the known results from linear algebra as well as the sparsity of the system. The proposed algorithms are proven to converge, and adaptive sparse channel estimation is used to demonstrate the effectiveness of the proposed approach.

Towards the main sequence: detailed analysis of weak line and post-T Tauri stars

A detailed study was performed for a sample of low-mass pre-main-sequence (PMS) stars, previously identified as weak-line T Tauri stars, which are compared to members of the Tucanae and Horologium Associations. Aiming to verify if there is any pattern of abundances when comparing the young stars at different phases, we selected objects in the range from 1 to 100 Myr, which covers most of PMS evolution. High-resolution optical spectra were acquired at European Southern Observatory and Observatorio do Pico dos Dias. The stellar fundamental parameters effective temperature and gravity were calculated by excitation and ionization equilibria of iron absorption lines. Chemical abundances were obtained via equivalent width calculations and spectral synthesis for 44 per cent of the sample, which shows metallicities within 0.5 dex solar. A classification was developed based on equivalent width of Li I 6708 angstrom and Ha lines and spectral types of the studied stars. This classification allowed a separation of the sample into categories that correspond to different evolutive stages in the PMS. The position of these stars in the Hertzsprung-Russell diagram was also inspected in order to estimate their ages and masses. Among the studied objects, it was verified that our sample actually contains seven weak-line T Tauri stars, three are Classical T Tauri, 12 are Fe/Ge PMS stars and 21 are post-T Tauri or young main-sequence stars. An estimation of circumstellar luminosity was obtained using a disc model to reproduce the observed spectral energy distribution. Most of the stars show low levels of circumstellar emission, corresponding to less than 30 per cent of the total emission.

SIMULATIONS OF WINDS OF WEAK-LINED T TAURI STARS. II. THE EFFECTS OF A TILTED MAGNETOSPHERE AND PLANETARY INTERACTIONS

Based on our previous work, we investigate here the effects on the wind and magnetospheric structures of weak-lined T Tauri stars due to a misalignment between the axis of rotation of the star and its magnetic dipole moment vector. In such a configuration, the system loses the axisymmetry presented in the aligned case, requiring a fully three-dimensional (3D) approach. We perform 3D numerical magnetohydrodynamic simulations of stellar winds and study the effects caused by different model parameters, namely the misalignment angle theta(t), the stellar period of rotation, the plasma-beta, and the heating index.. Our simulations take into account the interplay between the wind and the stellar magnetic field during the time evolution. The system reaches a periodic behavior with the same rotational period of the star. We show that the magnetic field lines present an oscillatory pattern. Furthermore, we obtain that by increasing theta(t), the wind velocity increases, especially in the case of strong magnetic field and relatively rapid stellar rotation. Our 3D, time-dependent wind models allow us to study the interaction of a magnetized wind with a magnetized extrasolar planet. Such interaction gives rise to reconnection, generating electrons that propagate along the planet`s magnetic field lines and produce electron cyclotron radiation at radio wavelengths. The power released in the interaction depends on the planet`s magnetic field intensity, its orbital radius, and on the stellar wind local characteristics. We find that a close-in Jupiter-like planet orbiting at 0.05 AU presents a radio power that is similar to 5 orders of magnitude larger than the one observed in Jupiter, which suggests that the stellar wind from a young star has the potential to generate strong planetary radio emission that could be detected in the near future with LOFAR. This radio power varies according to the phase of rotation of the star. For three selected simulations, we find a variation of the radio power of a factor 1.3-3.7, depending on theta(t). Moreover, we extend the investigation done in Vidotto et al. and analyze whether winds from misaligned stellar magnetospheres could cause a significant effect on planetary migration. Compared to the aligned case, we show that the timescale tau(w) for an appreciable radial motion of the planet is shorter for larger misalignment angles. While for the aligned case tau(w) similar or equal to 100 Myr, for a stellar magnetosphere tilted by theta(t) = 30 degrees, tau(w) ranges from similar to 40 to 70 Myr for a planet located at a radius of 0.05 AU. Further reduction on tau(w) might occur for even larger misalignment angles and/or different wind parameters.

Alternative diagnostic diagrams and the `forgotten` population of weak line galaxies in the SDSS

A numerous population of weak line galaxies (WLGs) is often left out of statistical studies on emission-line galaxies (ELGs) due to the absence of an adequate classification scheme, since classical diagnostic diagrams, such as [O iii]/H beta versus [N ii]/H alpha (the BPT diagram), require the measurement of at least four emission lines. This paper aims to remedy this situation by transposing the usual divisory lines between star-forming (SF) galaxies and active galactic nuclei (AGN) hosts and between Seyferts and LINERs to diagrams that are more economical in terms of line quality requirements. By doing this, we rescue from the classification limbo a substantial number of sources and modify the global census of ELGs. More specifically, (1) we use the Sloan Digital Sky Survey Data Release 7 to constitute a suitable sample of 280 000 ELGs, one-third of which are WLGs. (2) Galaxies with strong emission lines are classified using the widely applied criteria of Kewley et al., Kauffmann et al. and Stasinska et al. to distinguish SF galaxies and AGN hosts and Kewley et al. to distinguish Seyferts from LINERs. (3) We transpose these classification schemes to alternative diagrams keeping [N ii]/H alpha as a horizontal axis, but replacing H beta by a stronger line (H alpha or [O ii]), or substituting the ionization-level sensitive [O iii]/H beta ratio with the equivalent width of H alpha (W(H alpha)). Optimized equations for the transposed divisory lines are provided. (4) We show that nothing significant is lost in the translation, but that the new diagrams allow one to classify up to 50 per cent more ELGs. (5) Introducing WLGs in the census of galaxies in the local Universe increases the proportion of metal-rich SF galaxies and especially LINERs. In the course of this analysis, we were led to make the following points. (i) The Kewley et al. BPT line for galaxy classification is generally ill-used. (ii) Replacing [O iii]/H beta by W(H alpha) in the classification introduces a change in the philosophy of the distinction between LINERs and Seyferts, but not in its results. Because the W(H alpha) versus [N ii]/H alpha diagram can be applied to the largest sample of ELGs without loss of discriminating power between Seyferts and LINERs, we recommend its use in further studies. (iii) The dichotomy between Seyferts and LINERs is washed out by WLGs in the BPT plane, but it subsists in other diagnostic diagrams. This suggests that the right wing in the BPT diagram is indeed populated by at least two classes, tentatively identified with bona fide AGN and `retired` galaxies that have stopped forming stars and are ionized by their old stellar populations.

SIMULATIONS OF WINDS OF WEAK-LINED T TAURI STARS: THE MAGNETIC FIELD GEOMETRY AND THE INFLUENCE OF THE WIND ON GIANT PLANET MIGRATION

By means of numerical simulations, we investigate magnetized stellar winds of pre-main-sequence stars. In particular, we analyze under which circumstances these stars will present elongated magnetic features (e.g., helmet streamers, slingshot prominences, etc). We focus on weak-lined T Tauri stars, as the presence of the tenuous accretion disk is not expected to have strong influence on the structure of the stellar wind. We show that the plasma-beta parameter (the ratio of thermal to magnetic energy densities) is a decisive factor in defining the magnetic configuration of the stellar wind. Using initial parameters within the observed range for these stars, we show that the coronal magnetic field configuration can vary between a dipole-like configuration and a configuration with strong collimated polar lines and closed streamers at the equator (multicomponent configuration for the magnetic field). We show that elongated magnetic features will only be present if the plasma-beta parameter at the coronal base is beta(0) << 1. Using our self-consistent three-dimensional magnetohydrodynamics model, we estimate for these stellar winds the timescale of planet migration due to drag forces exerted by the stellar wind on a hot-Jupiter. In contrast to the findings of Lovelace et al., who estimated such timescales using the Weber and Davis model, our model suggests that the stellar wind of these multicomponent coronae are not expected to have significant influence on hot-Jupiters migration. Further simulations are necessary to investigate this result under more intense surface magnetic field strengths (similar to 2-3 kG) and higher coronal base densities, as well as in a tilted stellar magnetosphere.